For multi-machine equivalent in transient stability analysis， the grouping of the disturbed generators is the most important issue that needs to be solved. This paper investigates the clustering strategy for the disturbed generators by the fuzzy clustering method， analyzes the key parameters of this method and introduces the statistics ratio index F to ensure the flexibility of clustering number of disturbed generators. The standard New England 10-generator power system is used to test the effectiveness of the relevant method， the voting clustering method and the fuzzy clustering method. The test results illustrate that the fuzzy clustering method is self-adaptive and it is very effective as well as credible in clustering disturbed generators for power system transient stability analysis.